• Database Research
• /
• v.34 no.3
• /
• pp.45-57
• /
• 2018

The drowsiness that occurs in the driving is a very dangerous driver condition that can be directly linked to a major accident. In order to prevent drowsiness, there are traditional drowsiness detection methods to grasp the driver's condition, but there is a limit to the generalized driver's condition recognition that reflects the individual characteristics of drivers. In recent years, deep learning based state recognition studies have been proposed to recognize drivers' condition. Deep learning has the advantage of extracting features from a non-human machine and deriving a more generalized recognition model. In this study, we propose a more accurate state recognition model than the existing deep learning method by learning image and PPG at the same time to grasp driver's condition. This paper confirms the effect of driver's image and PPG data on drowsiness detection and experiment to see if it improves the performance of learning model when used together. We confirmed the accuracy improvement of around 3% when using image and PPG together than using image alone. In addition, the multimodal deep learning based model that classifies the driver's condition into three categories showed a classification accuracy of 96%.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.9
• /
• pp.407-418
• /
• 2023

The Multi-Object Goal Visual Navigation(MultiOn) is a visual navigation task in which an agent must visit to multiple object goals in an unknown indoor environment in a given order. Existing models for the MultiOn task suffer from the limitation that they cannot utilize an integrated view of multimodal context because use only a unimodal context map. To overcome this limitation, in this paper, we propose a novel deep neural network-based agent model for MultiOn task. The proposed model, MCFMO, uses a multimodal context map, containing visual appearance features, semantic features of environmental objects, and goal object features. Moreover, the proposed model effectively fuses these three heterogeneous features into a global multimodal context map by using a point-wise convolutional neural network module. Lastly, the proposed model adopts an auxiliary task learning module to predict the observation status, goal direction and the goal distance, which can guide to learn the navigational policy efficiently. Conducting various quantitative and qualitative experiments using the Habitat-Matterport3D simulation environment and scene dataset, we demonstrate the superiority of the proposed model.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.660-662
• /
• 2023

3차원 포인트 클라우드의 의미적 분할은 환경을 구성하는 물체 단위로 포인트 클라우드를 분할하는 작업으로서, 환경의 3차원적 구성을 이해하고 환경과 상호작용에 필수적인 시각 지능을 요구한다. 본 논문에서는 포인트 클라우드에서 추출하는 3차원 기하학적 특징과 함께 멀티-뷰 영상에서 추출하는 2차원 시각적 특징들도 활용하는 새로운 3차원 포인트 클라우드 의미적 분할 모델 MFNet을 제안한다. 제안 모델은 서로 이질적인 2차원 시각적 특징과 3차원 기하학적 특징의 효과적인 융합을 위해, 새로운 중기 융합 전략과 멀티-모달 교차 주의집중을 이용한다. 본 논문에서는 ScanNetV2 벤치마크 데이터 집합을 이용한 다양한 실험들을 통해, 제안 모델 MFNet의 우수성을 입증한다.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.629-632
• /
• 2021

기계 번역(machine translation)은 자연 언어로 된 텍스트를 다른 언어로 자동 번역 하는 기술로, 최근에는 주로 신경망 기계 번역(Neural Machine Translation) 모델에 대한 연구가 진행되었다. 신경망 기계 번역은 일반적으로 자기회귀(autoregressive) 모델을 이용하며 기계 번역에서 좋은 성능을 보이지만, 병렬화할 수 없어 디코딩 속도가 느린 문제가 있다. 비자기회귀(non-autoregressive) 모델은 단어를 독립적으로 생성하며 병렬 계산이 가능해 자기회귀 모델에 비해 디코딩 속도가 상당히 빠른 장점이 있지만, 멀티모달리티(multimodality) 문제가 발생할 수 있다. 본 논문에서는 단어 정렬(word alignment)을 이용한 비자기회귀 신경망 기계 번역 모델을 제안하고, 제안한 모델을 한국어-영어 기계 번역에 적용하여 단어 정렬 정보가 어순이 다른 언어 간의 번역 성능 개선과 멀티모달리티 문제를 완화하는 데 도움이 됨을 보인다.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.12
• /
• pp.505-518
• /
• 2023

3D point cloud semantic segmentation is a computer vision task that involves dividing the point cloud into different objects and regions by predicting the class label of each point. Existing 3D semantic segmentation models have some limitations in performing sufficient fusion of multi-modal features while ensuring both characteristics of 2D visual features extracted from RGB images and 3D geometric features extracted from point cloud. Therefore, in this paper, we propose MMCA-Net, a novel 3D semantic segmentation model using 2D-3D multi-modal features. The proposed model effectively fuses two heterogeneous 2D visual features and 3D geometric features by using an intermediate fusion strategy and a multi-modal cross attention-based fusion operation. Also, the proposed model extracts context-rich 3D geometric features from input point cloud consisting of irregularly distributed points by adopting PTv2 as 3D geometric encoder. In this paper, we conducted both quantitative and qualitative experiments with the benchmark dataset, ScanNetv2 in order to analyze the performance of the proposed model. In terms of the metric mIoU, the proposed model showed a 9.2% performance improvement over the PTv2 model using only 3D geometric features, and a 12.12% performance improvement over the MVPNet model using 2D-3D multi-modal features. As a result, we proved the effectiveness and usefulness of the proposed model.

• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.278-284
• /
• 2009

Multimodal data to have several modalities such as videos, images, sounds and texts for one contents is increasing. Since this type of data has ill-defined format, it is not easy to represent the crossmodal information for them explicitly. So, we proposed new method to extract and analyze vision-language crossmodal association information using the documentaries video data about the nature. We collected pairs of images and captions from 3 genres of documentaries such as jungle, ocean and universe, and extracted a set of visual words and that of text words from them. We found out that two modal data have semantic association on crossmodal association information from this analysis.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.468-472
• /
• 2020

시각적 질의응답(Visual Question Answering, VQA)은 주어진 이미지에 연관된 다양한 질문에 대한 올바른 답변을 예측하는 기술이다. 해당 기술은 컴퓨터 비전-자연어 처리 연구분야에서 활발히 연구가 진행되고 있으며, 질문의 의도를 정확히 파악하고, 주어진 이미지에서 관련 단서 정보를 찾는 것이 중요하다. 또한, 서로 이질적인 특성을 지닌 정보(이미지 객체, 객체 위치, 질문)를 통합하는 과정도 중요하다. 본 논문은 질문의 의도에 알맞은 정보를 효율적으로 사용하기 위해 멀티 모달 입력 이미지 객체, 객체 위치, 질문)에 대한 Multi-modal Message Aggregation (MMA) 제안하며 이를 통해 한국어 시각적 질의응답 KVQA에서 다른 모델보다 더 좋은 성능을 확인하였다.

• The Journal of the Korea Contents Association
• /
• v.22 no.1
• /
• pp.138-147
• /
• 2022

Recently, chatbots have been applied in various fields and have shown good results, and many attempts to use chatbots in shopping mall product recommendation services are being conducted on e-commerce platforms. In this paper, for a conversation system that recommends a fashion that a user wants based on conversation between the user and the system and fashion image information, a transformer model that is currently performing well in various AI fields such as natural language processing, voice recognition, and image recognition. We propose a multimodal-based improved transformer model that is improved to increase the accuracy of recommendation by using dialogue (text) and fashion (image) information together for data preprocessing and data representation. We also propose a method to improve accuracy through data improvement by analyzing the data. The proposed system has a recommendation accuracy score of 0.6563 WKT (Weighted Kendall's tau), which significantly improved the existing system's 0.3372 WKT by 0.3191 WKT or more.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2006.06a
• /
• pp.57-62
• /
• 2006

휴대용 단말기의 소형화 및 지능화와 더불어 차세대 PC 기반의 유비쿼터스 컴퓨팅에 대한 관심이 높아짐에 따라 최근에는 펜이나 음성 입력 멀티미디어 등 여러 가지 대화 모드를 구비한 멀티 모달 상호작용 (Multi-Modal Interaction MMI)에 대한 연구가 활발히 진행되고 있다. 따라서, 본 논문에서는 잡음 환경에서의 명확한 의사 전달 및 휴대용 단말기에서의 음성-제스처 통합 인식을 위한 인터페이스의 연구를 목적으로 Voice-XML과 Wearable Personal Station(WPS) 기반의 음성 및 내장형 수화 인식기를 통합한 멀티 모달 명령어 인식 시스템 (Multi-Modal Instruction Recognition System : MMIRS)을 제안하고 구현한다. 제안되어진 MMIRS는 한국 표준 수화 (The Korean Standard Sign Language : KSSL)에 상응하는 문장 및 단어 단위의 명령어 인식 모델에 대하여 음성뿐만 아니라 화자의 수화제스처 명령어를 함께 인식하고 사용함에 따라 잡음 환경에서도 규정된 명령어 모델에 대한 인식 성능의 향상을 기대할 수 있다. MMIRS의 인식 성능을 평가하기 위하여, 15인의 피험자가 62개의 문장형 인식 모델과 104개의 단어인식 모델에 대하여 음성과 수화 제스처를 연속적으로 표현하고, 이를 인식함에 있어 개별 명령어 인식기 및 MMIRS의 평균 인식율을 비교하고 분석하였으며 MMIRS는 문장형 명령어 인식모델에 대하여 잡음환경에서는 93.45%, 비잡음환경에서는 95.26%의 평균 인식율을 나타내었다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.1
• /
• pp.27-32
• /
• 2019

Research that implements the classification process through Deep Learning algorithm, one of the outstanding human abilities, includes a unimodal model, a multi-modal model, and a multi-modal method using music videos. In this study, the results were better by suggesting a system to analyze each song's spectrum into short samples and vote for the results. Among Deep Learning algorithms, CNN showed superior performance in the category of music genre compared to RNN, and improved performance when CNN and RNN were applied together. The system of voting for each CNN result by Deep Learning a short sample of music showed better results than the previous model and the model with Softmax layer added to the model performed best. The need for the explosive growth of digital media and the automatic classification of music genres in numerous streaming services is increasing. Future research will need to reduce the proportion of undifferentiated songs and develop algorithms for the last category classification of undivided songs.